DNA Nanostar System: Programmable Soft Matter

• DNA nanostar systems are programmable soft-matter constructs composed of multi-arm DNA junctions that self-assemble via sticky ends to form equilibrium gels, droplets, and clusters.
• They rely on design parameters such as valence, arm length, sticky-end sequences, and junction flexibility to control phase behavior, rheology, and emergent mechanics.
• These systems have practical applications in responsive hydrogels, drug delivery, biosensing, and engineered scaffolds due to their tunable assembly and dynamic properties.

Searching arXiv for recent and foundational papers on DNA nanostar systems to ground the encyclopedia entry. arXiv search query: DNA nanostar system phase separation hydrogel rheology surfactant site:arxiv.org DNA nanostar systems are programmable soft-matter systems built from self-assembled branched DNA motifs whose finite valence, sticky-end sequence design, junction flexibility, and environmental responsiveness permit controlled formation of equilibrium gels, transient hydrogels, liquid–liquid phase-separated droplets, hyperbranched clusters, and enzyme-responsive scaffolds. Across the literature, “DNA nanostar” denotes a multi-arm DNA junction in which several single-stranded oligonucleotides hybridize into a central core with double-stranded arms terminating in single-stranded sticky ends; these sticky ends mediate reversible, sequence-specific interparticle bonding and thereby connect nanostar architecture to mesoscale thermodynamics, mechanics, and transport (Rovigatti et al., 2014). Recent work has expanded this concept from single-component tetravalent gels to phase-separated droplet ensembles with engineered interfacial properties (Gao et al., 2023), multi-phase condensate mixtures with up to 9 distinct non-adhering phases (Chaderjian et al., 26 Aug 2025), power-law rheological networks (Conrad et al., 2023), and programmable enzymatic degradation and antibody release in hydrogel scaffolds (Palombo et al., 21 Jan 2026).

1. Molecular architecture and design space

DNA nanostars are defined by arm number, core flexibility, arm length, sticky-end identity, and the presence or absence of auxiliary elements such as linkers or long DNA surfactants. Multiple studies use tetravalent nanostars assembled from four single-stranded oligonucleotides into a four-way junction with four double-stranded arms, often ending in palindromic sticky ends such as $5'$-CGATCG-$3'$, so that any arm can bind any other arm of the same species (Gao et al., 2023). Other work focuses on trivalent Y-shaped nanostars assembled from three oligonucleotides, with each arm carrying one sticky end or, in linker-mediated systems, binding a double-stranded linker that mediates inter-nanostar crosslinking (Palombo et al., 21 Jan 2026). Six-arm designs have also been used to create transient hydrogels with coexisting strong and weak bond classes, enabling controlled power-law rheology (Conrad et al., 2023).

The central junction is a major determinant of mechanics and assembly. In several designs, unpaired adenines or thymines are inserted at the core and near the sticky ends, producing flexible joints that modulate angular freedom, steric accessibility, and effective rigidity (Fosado, 2022). In tetravalent nanostars, low-salt oxDNA simulations indicate a predominantly planar “X-shaped” conformation rather than a rigid tetrahedral one, with anisotropy closer to a planar object than to a perfect tetrahedral star (Rovigatti et al., 2014). In trivalent nanostars, planarity can be quantified by a collective variable $d_p$, whose free-energy landscape shifts from planar to non-planar configurations as the number of unpaired bases at the core increases or salt concentration rises (Fosado, 2022). This establishes geometry as a state variable rather than a fixed structural descriptor.

Sticky-end design encodes both bond thermodynamics and interaction topology. Palindromic sticky ends support homotypic self-binding and are widely used in one-component droplet and gel systems (Wilken et al., 2022). Orthogonal self-complementary sticky ends allow the coexistence of multiple immiscible droplet species in the same sample (Wilken et al., 2022). More generally, palindromic sticky-end design can be treated combinatorially: for even sticky-end length $M$, the number of possible palindromes is $4^{M/2}$, but orthogonality constraints and free-energy matching sharply reduce the usable set for multi-phase design (Chaderjian et al., 26 Aug 2025). In one experimentally realized case, graph-theoretic selection under 3-base orthogonality and $|\Delta G_\alpha-\Delta G_\beta|<\sigma_G$ yielded a 9-member clique of mutually orthogonal sticky ends, enabling 9 distinct non-adhering droplet phases (Chaderjian et al., 26 Aug 2025).

A useful implication is that the “DNA nanostar system” is not a single material class but a parameterized family of limited-valence associating particles. Valence, core flexibility, arm length, linker geometry, and sticky-end sequence jointly determine whether the dominant state is a dilute fluid, a percolating equilibrium gel, a cluster fluid, a phase-separated droplet liquid, or an enzyme-responsive hydrogel (Stoev et al., 2019).

2. Thermodynamics of association, gelation, and phase separation

The central thermodynamic mechanism in DNA nanostar systems is reversible sticky-end hybridization. Association is driven by the enthalpic gain of Watson–Crick base pairing and base stacking, opposed by the loss of translational, rotational, and configurational entropy when nanostars are confined into clusters, networks, or dense phases (Gao et al., 2023). Salt screens the electrostatic repulsion of DNA backbones and thereby stabilizes both nanostar formation and inter-nanostar bonding (Rovigatti et al., 2014).

A widely used theoretical description treats nanostars as patchy particles within Wertheim’s theory of associating fluids. For tetravalent nanostars, the Helmholtz free energy is written as a reference part plus a bonding contribution, with the latter expressed through the sticky-end bond probability $p_b$ (Rovigatti et al., 2014). A phase-field implementation based on the Cahn–Hilliard equation adopts a free-energy functional

$$F(T,V)=\int_V \left[f(\{\rho_i(\mathbf r)\})+\frac{K}{2}\sum_i(\nabla \rho_i)^2\right]\,dV,$$

where $f=f_{\text{ref}}+f_{\text{bond}}$ is derived from Wertheim theory and nearest-neighbor DNA thermodynamics (Cappa et al., 8 Jan 2025). In the one-component case, the fraction of unbonded sticky ends is

$$X_\alpha(T,\rho_A,[Na^+])=\frac{-1+\sqrt{1+4\Delta_\alpha \mathcal V \rho_A}}{2\Delta_\alpha \mathcal V \rho_A},$$

with $3'$0 for tetravalent nanostars and $3'$1 set by sticky-end hybridization free energy (Cappa et al., 8 Jan 2025). This construction reproduces gas–liquid coexistence, spinodal decomposition, surface-tension variation, and multicomponent morphology in DNA-made particle solutions (Cappa et al., 8 Jan 2025).

For tetravalent nanostars at $3'$2 M NaCl in oxDNA simulations, the phase behavior shows a gas–liquid phase-separation region and an equilibrium gel regime at higher concentrations (Rovigatti et al., 2014). The critical temperature is around $3'$3C, with critical concentration near $3'$4 mg/ml in simulation units adjusted for salt, and the critical bond probability is about $3'$5 (Rovigatti et al., 2014). Percolation occurs when the fraction of bonded arms exceeds a concentration-dependent threshold corresponding to $3'$6, larger than mean-field $3'$7 because loops and double bonds reduce effective connectivity (Rovigatti et al., 2014).

The same thermodynamic logic underlies liquid–liquid phase separation in micron-scale droplet systems. Tetravalent nanostars with short sticky ends at high salt separate into a dense nanostar-rich droplet phase and a dilute phase below a critical temperature, yielding an equilibrium model condensate (Gao et al., 2023). In another tetravalent system, phase separation proceeds via a binodal transition and can be modeled by the Cahn–Hilliard equation

$3'$8

with near-equilibrium droplet organization governed by subsequent Brownian diffusion (Wilken et al., 2022).

3. Network topology, geometry, and emergent mechanics

The macroscopic mechanics of DNA nanostar systems depend not only on valence but also on nanostar geometry, loop formation, linker flexibility, and the coexistence of multiple bond classes. Tetravalent equilibrium gels at high enough concentration exhibit progressive growth of an interparticle structure-factor peak upon cooling, with the effective structure factor $3'$9 extracted from

$d_p$0

where $d_p$1 is the temperature-independent nanostar form factor obtained from coarse-grained simulations mapped to pseudo-atomistic coordinates (Fernandez-Castanon et al., 2016). This separation of form factor and structure factor shows that cooling drives bond formation and network development rather than major monomer shape changes (Fernandez-Castanon et al., 2016).

In trivalent nanostar hydrogels, planarity modulates mechanics strongly. A multiscale simulation framework combining oxDNA, metadynamics, and a coarse-grained seven-bead nanostar model showed that non-planar nanostars have higher bond saturation but lower zero-shear viscosity than planar nanostars; specifically, viscosities differed by about a factor of 20 at equal volume fraction and interaction strength (Fosado, 2022). The zero-shear viscosity was obtained from the Green–Kubo relation

$d_p$2

and the lower viscosity of non-planar systems was linked to easier local rearrangements and different cluster topology (Fosado, 2022). This suggests that microscopic geometry can be as important as valence in controlling gel rheology.

Linker-mediated trivalent hydrogels reveal a different aspect of topology. When Y-shaped nanostars are connected through rigid linear linkers, the system undergoes classical gelation on cooling, with DLS microrheology showing a transition from fluid to elastic response (Stoev et al., 2019). When the linker contains sufficiently long non-binding thymine joints, however, a single linker can bridge two arms of the same nanostar, producing loops and reducing effective valence below the percolation threshold. The result is a cluster fluid with reduced viscosity rather than a percolating gel (Stoev et al., 2019). This is a direct demonstration that local loop topology can override nominal arm count.

Loop formation is even more explicit in AB$d_p$3 tetravalent nanostars designed with one A sticky end and three B sticky ends, where only AB bonds are allowed. Contrary to the standard Flory–Stockmayer prediction that full bond conversion should imply percolation, oxDNA simulations and DLS experiments show that the fully bonded state remains a sol of finite hyperbranched clusters because the final unreacted A often forms an intracluster loop rather than an intercluster bridge (Lattuada et al., 2020). The cluster size distribution is described by an extended Flory–Stockmayer framework,

$d_p$4

where $d_p$5 and $d_p$6 are partition functions for loopless and looped clusters, respectively (Lattuada et al., 2020). In the low-temperature fully bonded limit, the distribution remains finite and concentration dependent, establishing that full conversion does not imply network percolation when local cyclization is entropically favorable (Lattuada et al., 2020).

Rheologically, the most detailed sequence-level control has been achieved in 6-arm nanostar hydrogels with weak $d_p$7 and strong $d_p$8 bonds. Homotypic gels with a single bond class show Maxwell-like viscoelasticity, while heterotypic gels with mixed strong and weak bonds exhibit power-law frequency dependence of $d_p$9 and $M$0 (Conrad et al., 2023). A diffusive stress-relaxation model posits that the strong-bond sub-network relaxes by diffusing through an effective viscosity set by the weak bonds. In the Zimm limit,

$M$1

and the storage modulus scales as

$M$2

where $M$3 is the fractal dimension of the shortest stress-bearing path (Conrad et al., 2023). In the Rouse limit,

$M$4

with $M$5 the fractal dimension of the strong-bond network (Conrad et al., 2023). Comparison to data implies Zimm-like behavior for $M$6 and $M$7, but Rouse-like behavior for $M$8, indicating that the density of weak bonds controls the hydrodynamic coupling regime (Conrad et al., 2023).

A related but distinct rheological phenomenon has been observed in a simple 3-arm hydrogel with 16 bp arms and 6 bp sticky ends. Below sticky-end percolation, bulk rheology shows systematic deviation from classical Maxwell behavior and is better captured by a fractional Maxwell model with temperature-independent exponents $M$9 and $4^{M/2}$0, indicating a broad relaxation spectrum associated with structural disorder (Ajiyel et al., 20 Mar 2026). At temperatures much higher than the percolation threshold, DLS microrheology reveals an elastic plateau interpreted as glassy arrest of crowded nanostar doublets rather than network formation, with a calculated effective packing fraction $4^{M/2}$1 at the onset of solidity (Ajiyel et al., 20 Mar 2026). This suggests that disorder and crowding can produce solid-like mechanics even without a percolated bond network.

4. Condensates, droplets, and interfacial engineering

DNA nanostar systems have become a model platform for biomolecular condensation because their droplets are sequence-programmable, liquid-like, and structurally tractable. Tetravalent nanostars with short palindromic sticky ends in 1 M NaCl form micron-scale liquid droplets on cooling after homogenization at $4^{M/2}$2C (Gao et al., 2023). These droplets have an extremely low interfacial tension, around $4^{M/2}$3N/m, so coarsening is dominated by coalescence rather than Ostwald ripening (Gao et al., 2023). Their interiors are dense and nearly fully bonded, but the sticky-end bonds remain dynamic, making the dense phase a disordered, dynamic mesh that behaves like a liquid on long timescales (Gao et al., 2023).

One major development is interfacial control via surface-active DNA. Long 401 bp double-stranded DNA molecules terminated with a sticky end complementary to the nanostar sticky sequence partition preferentially to the droplet interface because binding to surface nanostars is favorable while deep penetration into the dense phase is entropically costly (Gao et al., 2023). Confocal imaging and point-spread-function-based analysis indicate an interfacial density of about 20 surfactant molecules per $4^{M/2}$4, corresponding to a mean spacing of roughly $4^{M/2}$5 nm and a sparse brush-like layer (Gao et al., 2023). This sparse polymer brush generates steric disjoining pressure upon droplet approach, arresting coalescence and suppressing adhesion to solids. As surfactant concentration increases, droplets shrink from $4^{M/2}$6m scale to sub-micron sizes and eventually to $4^{M/2}$7 nm-scale assemblies observed by DLS and TEM, consistent with swollen micelle-like nanostar–surfactant particles (Gao et al., 2023). The same surfactant layer also prevents adhesion to hydrophobic and BSA-coated hydrophilic surfaces, apparently by keeping nanostar sticky ends from making direct contact (Gao et al., 2023).

Another major theme is spatial organization. Sedimented phase-separated tetravalent nanostar droplets spontaneously form hyperuniform patterns, diagnosed from image-based structure factors $4^{M/2}$8 with small-$4^{M/2}$9 scaling

$|\Delta G_\alpha-\Delta G_\beta|<\sigma_G$0

(Wilken et al., 2022). This $|\Delta G_\alpha-\Delta G_\beta|<\sigma_G$1 hyperuniformity is interpreted as the remnant of an initial Cahn–Hilliard pattern softened by Brownian diffusion of droplets over approximately one inter-droplet spacing (Wilken et al., 2022). A phenomenological fit,

$|\Delta G_\alpha-\Delta G_\beta|<\sigma_G$2

introduces a hyperuniformity length scale $|\Delta G_\alpha-\Delta G_\beta|<\sigma_G$3 at fixed time, consistent with Stokes–Einstein diffusion of droplets (Wilken et al., 2022). Two-species systems with orthogonal sticky ends produce two independent hyperuniform droplet fields in the same sample, while the combined pattern is not hyperuniform because cross-species correlations are random, arguing against hydrodynamic interactions as the primary ordering mechanism (Wilken et al., 2022).

Phase diversity has recently been pushed much further. By designing 9 sticky-end palindromes satisfying stringent orthogonality and near-matched $|\Delta G_\alpha-\Delta G_\beta|<\sigma_G$4, one study demonstrated 9 distinct, non-adhering nanostar phases that do not share components (Chaderjian et al., 26 Aug 2025). Fluorescence-encoded imaging distinguished the phases, and “single-dark” control experiments showed partition coefficients below 1 for all cross-phase combinations, implying less than 1 heterotypic nanostar per $|\Delta G_\alpha-\Delta G_\beta|<\sigma_G$5 homotypic nanostars in the dense phase (Chaderjian et al., 26 Aug 2025). Rapid quenching of 7 such species produced a densely packed 2D droplet layer with glassy dynamics, slow coarsening, and dynamic heterogeneity due to caging by unlike droplets (Chaderjian et al., 26 Aug 2025). This suggests that sequence-encoded interaction matrices can generate not only independent phases but collective glassy states in multi-condensate systems.

Electrokinetic nanofluidics provides an orthogonal route to interrogate condensates. Four-armed nanostar condensates formed in silica nanochannels alter ionic current, allowing electronic detection of the phase transition (Chou et al., 2024). The condensate electrophoretic mobility is about half that of a single nanostar, and its effective ionic strength is about 35% greater than that of 150 mM NaCl in phosphate buffer (Chou et al., 2024). Zeta-potential measurements before and after exposure to nanostars indicate that condensate binds silica walls more strongly than isolated nanostars, increasing electroosmotic flow enough to balance or exceed the electrophoretic motion of condensate (Chou et al., 2024). A tilted-channel configuration yields robust electronic signatures of condensation temperatures $|\Delta G_\alpha-\Delta G_\beta|<\sigma_G$6 matching those obtained by other methods (Chou et al., 2024). A plausible implication is that DNA nanostar condensation can be used as an electronically readable state variable in biosensing platforms.

5. Surface binding, selective recognition, and reconfiguration

Beyond bulk assembly and phase separation, DNA nanostar systems are also used to study multivalent surface interactions. In one experimental model, nanostars of valency $|\Delta G_\alpha-\Delta G_\beta|<\sigma_G$7 bind to complementary DNA receptors mobile in supported lipid bilayers, allowing direct measurement of superselectivity by TIRF microscopy (Linne et al., 2023). The bound fraction $|\Delta G_\alpha-\Delta G_\beta|<\sigma_G$8 is analyzed through the selectivity exponent

$|\Delta G_\alpha-\Delta G_\beta|<\sigma_G$9

where $p_b$0 is receptor density (Linne et al., 2023). Monovalent nanostars follow Langmuir behavior,

$p_b$1

while multivalent nanostars require an avidity-based description with an effective intra-particle association constant $p_b$2 (Linne et al., 2023). Experimentally, trivalent nanostars display the highest maximal superselectivity, despite the classical expectation that selectivity might grow with valence (Linne et al., 2023). To explain this, the theory was extended to include pair interactions between simultaneously bound arms through an effective $p_b$3, with fitted values $p_b$4 for trivalent stars, $p_b$5 for hexavalent, and $p_b$6 for 10-arm stars (Linne et al., 2023). This indicates that weak cooperativity favors three-arm designs, whereas higher-valence stars are penalized by competitive interactions, receptor depletion, and steric crowding (Linne et al., 2023).

The same programmable binding logic can be extended to structural trans-assembly. Although not itself a nanostar study, a strand-displacement framework for DNA “trans-assembly” is directly adaptable to nanostars: one DNA structure releases a signal strand that triggers a circuit, which then activates precursor monomers of another structure (Shin et al., 2018). The data suggest that nanostars could be equipped with hairpin-encoded signals or protected sticky ends, enabling one nanostar assembly to trigger formation of another topology through toehold-mediated circuits (Shin et al., 2018). This suggests a route to state-dependent nanostar reconfiguration, though the specific nanostar implementation remains prospective.

6. Responsive hydrogels, degradation, and applications

DNA nanostar hydrogels are increasingly viewed as programmable biomaterials rather than only model patchy colloids. A recent study of three-armed DNA nanostar hydrogels examined how flexible joints, arm lengths, mesh sizes, and the placement of restriction-enzyme sites govern degradability and cargo release (Palombo et al., 21 Jan 2026). In linker-free design A, sticky ends are self-complementary 6-nt sequences, yielding a directly crosslinked network. In designs B–D, nanostar arms carry 8-nt non-palindromic sticky ends that bind double-stranded linkers, increasing core-to-core spacing to about $p_b$7 nm (Palombo et al., 21 Jan 2026). Flexible joints are implemented as unpaired adenines at the core (FJ1) and near sticky ends (FJ2), modulating both network mechanics and enzyme accessibility (Palombo et al., 21 Jan 2026).

Microrheology tracks degradation via tracer-particle mean-squared displacement,

$p_b$8

the diffusion coefficient

$p_b$9

and the effective viscosity

$$F(T,V)=\int_V \left[f(\{\rho_i(\mathbf r)\})+\frac{K}{2}\sum_i(\nabla \rho_i)^2\right]\,dV,$$0

with normalized viscosity $$F(T,V)=\int_V \left[f(\{\rho_i(\mathbf r)\})+\frac{K}{2}\sum_i(\nabla \rho_i)^2\right]\,dV,$$1 used to follow softening (Palombo et al., 21 Jan 2026). Site-specific restriction enzymes such as EcoRI, EcoRV, BtgI, and MspA1I are often ineffective when their sites are embedded on nanostar arms near crowded cores, whereas DNase I rapidly degrades the hydrogels regardless of design (Palombo et al., 21 Jan 2026). Crucially, moving a restriction site from an arm to a more exposed linker strongly enhances cleavage, and longer-arm design D undergoes a much larger viscosity drop than design B under identical EcoRV conditions, with $$F(T,V)=\int_V \left[f(\{\rho_i(\mathbf r)\})+\frac{K}{2}\sum_i(\nabla \rho_i)^2\right]\,dV,$$2 for D versus $$F(T,V)=\int_V \left[f(\{\rho_i(\mathbf r)\})+\frac{K}{2}\sum_i(\nabla \rho_i)^2\right]\,dV,$$3 for B (Palombo et al., 21 Jan 2026). This shows that geometry and site placement jointly control degradability.

The same design logic can be used for cargo release. Design D, with relatively large pores around 34 nm, can entrap bevacizumab antibodies and retain them almost completely in the absence of enzymatic degradation, with passive release below 2 ng/mL after 48 h (Palombo et al., 21 Jan 2026). EcoRV cleavage of linker sites then drives gradual release to about 64 ng/mL by 73 h, whereas DNase I, effective in pure gels, is surprisingly suppressed in the presence of antibody, possibly due to antibody–DNA interactions that inhibit enzyme access (Palombo et al., 21 Jan 2026). This establishes programmable degradation as a functional handle for responsive drug-delivery materials.

More broadly, DNA nanostar systems are repeatedly positioned as model systems for patchy colloids, biomolecular condensates, and synthetic cells. Their advantages include well-defined valence, tunable bond free energies, orthogonal interactions, reversible assembly, and compatibility with microscopy, scattering, rheology, and nanofluidics (Fernandez-Castanon et al., 2016). Applications discussed across the literature include biosensing, drug and antibody delivery, tissue scaffolds, synthetic organelles, hyperuniform biochemical reactor arrays, and optical or mechanical metamaterials (Wilken et al., 2022). The claim that they are useful model condensates is particularly strong because surface tension, network connectivity, and interfacial composition can all be tuned independently enough to test hypotheses that are difficult to isolate in protein–RNA condensates (Gao et al., 2023).

A recurring misconception is that DNA nanostar materials are necessarily highly ordered because DNA nanotechnology is often associated with precise origami and crystals. The evidence instead shows that many nanostar systems are intentionally disordered, and that this disorder is essential: it permits equilibrium gelation without crystallization (Fernandez-Castanon et al., 2016), broad power-law relaxation spectra (Conrad et al., 2023), loop-stabilized hyperbranched sols (Lattuada et al., 2020), and glass-like solidity in non-percolated states (Ajiyel et al., 20 Mar 2026). Another misconception is that valence alone determines mechanics. Geometry, flexibility, loop formation, and interfacial composition repeatedly emerge as equally important control parameters (Fosado, 2022).

Taken together, the DNA nanostar system is best understood as a unifying framework for sequence-programmable limited-valence matter. It spans associating fluids, gels, clusters, condensates, and responsive hydrogels, with microscopic control over geometry and bond thermodynamics translating into predictable macroscopic changes in phase behavior, rheology, and interfacial function (Rovigatti et al., 2014). This suggests that future progress will likely come from combining the established handles—valence, planarity, bond heterogeneity, orthogonality, linker placement, surfactant layers, and enzymatic addressability—into fully multiscale design rules for DNA-based soft materials (Cappa et al., 8 Jan 2025).